/AI4h ago

York University's Kosta Derpanis proposes making open code and weights mandatory for computer vision paper awards

Industry-aligned researchers recently pitched a similar motion to PAMITC.

563234.3K
#932Kosta Derpanis (sabbatical in Zurich)
Andrea Tagliasacchi @CVPR
Original post
Kosta Derpanis (sabbatical in Zurich)@CSProfKGD#932inAI

#KostasThoughts: I’ve been thinking about the criteria for award-winning papers 🏆 They should embody what our community values most: originality, potential impact, and reproducibility. Yet (full) reproducibility is often missing, even among award-winning papers.

Making code and model weights publicly available should be a criterion for award consideration. This would send a strong signal to decision-makers in both academia and industry about the importance of open and reproducible research.

Papers would receive the recognition they deserve, while the community would gain a deeper understanding of the engineering innovations that drive performance.

Thoughts?

8:38 AM · Jun 8, 2026 · 4K Views
Sentiment

Positive users agree reproducibility rules for award-winning papers could enable automatic verification of agent-based results, while negative users warn specialized industry code would prevent meaningful reproduction by others.

Pos
50.0%
Neg
50.0%
2 comments with sentiment.
Cluster Engagement
Posts from X
Most Activity
Most Activity
VIEWS386BOOKMARKS1LIKES3
Andrea Tagliasacchi @CVPR@taiyasaki

@CSProfKGD This is the argument Marc also made after I pitched the motion at the PAMITC

What is critical to note is that all people that spoke up there had industrial affiliations: Georgia, Marc, Myself.

This is not even something a group of pure academics requested

Kosta Derpanis (sabbatical in Zurich)@CSProfKGD

#KostasThoughts: I’ve been thinking about the criteria for award-winning papers 🏆 They should embody what our community values most: originality, potential impact, and reproducibility. Yet (full) reproducibility is often missing, even among award-winning papers.

Making code and model weights publicly available should be a criterion for award consideration. This would send a strong signal to decision-makers in both academia and industry about the importance of open and reproducible research.

Papers would receive the recognition they deserve, while the community would gain a deeper understanding of the engineering innovations that drive performance.

Thoughts?

2hViews 386Likes 3Bookmarks 1
Jeffrey Li 💙💛@askerlee

@CSProfKGD Yeah esp. in the age of agents, the experimental results can often be automatically verified if the compute budget is not too high

4hViews 187Likes 2Bookmarks 1
Zeeshan Zia@ZeeshanZiaML

One challenge is that code at big companies is too specialized for their infrastructure. Releasing it doesn't enable the broader community to understand it, nor can the community frankly reproduce something trained on 1000s of GPUs.

The whole community will be poorer if they discourage industrial labs to publish, at a time, when the most impactful work is being done in industry!

2hViews 2